186 JV. L. Boioen — Crystallization — 



l/3 cm per hr. These examples agree in placing the absolute 

 viscosity of the mixtures at roughly 25 times that of the less 

 siliceous mixtures or at about 100. 



The still more siliceous mixture which exhibited the floating 

 of tridymite crystals shows a further increase of viscosity. 

 The difference in density between tridymite and the liquid is 

 approximately the same as that between forsterite and the 

 liquid from which it settled, but of the opposite sign. Crystals 

 of tridymite and of forsterite of the same size should rise in 

 the one case or sink in the other at approximately the same 

 rate in liquids of the same viscosity. But crystals of tridymite 

 rise in the artificial melt at about 1/50 the rate at which fors- 

 terite crystals of the same size sink. The viscosity of the 

 liquid is therefore approximately 50 times as great, or in abso- 

 lute units about 200. 



The liquid* of composition, pyroxene 98 per cent, Si0 2 2 

 per cent, has a viscosity of about 4 at 1430°, pyroxene 91 per 

 cent, Si0 2 9 per cent at the same temperature, a viscosity of 

 about 100, and pyroxene 86*5 percent, Si0 2 13*5 per cent, a vis- 

 cosity of about 200. All of these are rough estimates and tend 

 to be rather too large than too small. 



Differentiation. 



When the crystallization of the liquid is completed in those 

 artificial preparations in which olivine crystals sank, the upper 

 part consists of pyroxene and free silica and the lower part of 

 pyroxene and olivine. It is sometimes suggested that the sinking 

 (or rising) of minerals in the magma has been a factor in pro- 

 ducing differentiation in cases where the differentiates consist 

 of the same minerals in varying proportions. Thus Ussing 

 thinks the process likely for his nephelite-sodalite-eudialyte 

 syenites, which are rich in sodalite in the upper layers and rich 

 in eudialyte and other heavy minerals in the lower layers ; but 

 he denies its general importance.f The result of the sinking 

 of crystals is, however, not necessarily of such a simple char- 

 acter. A partially different assemblage of minerals may result, 

 as shown in the comparatively simple artificial mixtures de- 

 scribed above, and the likelihood of such phenomena is greater 

 in more complex systems. 



The Sinking of Crystals in the Palisade Diabase Magma. 

 J. V. Lewis describes an olivine-rich layer in the Palisade 

 diabase as the olivine-diabase ledge. It occurs 40-50 feet 



* The composition of the liquid is obtained by correcting the total com- 

 position of the original mixture by the amount necessary on account of the 

 separation from it of the small proportion of crystals. 



f Geology of the Country around Julianehaab, Greenland, p. 347. 



